Claude Code release notes: 2.1.312 adds cloud routine effort

Claude Code 2.1.312 adds an effort setting to the model picker when you create or edit a routine that runs in the cloud. Effort controls how much reasoning work the model puts into a task.

Claude Code: choose effort for a cloud routine

Human step: When creating or editing a routine that runs in the cloud, use the model picker to choose effort. The changelog does not list the available levels for each model or announce new availability by subscription plan.

The same release also fixes Edit in a routine's menu on the Routines page. Previously, this could open the routine's page without opening its editing dialog.

Source: Anthropic's Claude Code 2.1.312 release notes, Cloud sessions.

Clearer feedback when effort changes during a task

A newly selected effort level may take effect later during an ongoing task. Claude Code now corrects the spinner, transcripts and OpenTelemetry effort values, used for operational monitoring, to show the level actually in use rather than the newly selected level.

The prompt's effort indicator now says a change may wait for your next message. The note clears when it no longer applies. The /model and /effort cache warning is also corrected: a change made while Claude is working can trigger a prompt re-read before your next message.

Source: Claude Code 2.1.312: effort indicator and cache warning.

GitHub published version 2.1.312 on October 27. When checked on October 29, it was also the latest version in the npm registry.

Source: Anthropic's Claude Code package in the npm registry.